Hydro-geomorphic response models for burned areas and their applications in land management

Abstract

Erosion, flash floods and debris flows are hydro-geomorphic processes that intensify due to catchment disturbance by wildland fire. Predictive models of these processes are used by land managers to quantify rehabilitation effectiveness, prioritize resources and evaluate trade-offs between different management strategies. Predictions can be difficult to make, however, because of heterogeneous landscapes, stochastic rainfall, and the transient and variable fire effects. This paper reviews hydro-geomorphic response models for burned areas and explores how modelling approaches and sources of uncertainty change depending on the focus question (or purpose) and the associated spatial-temporal scale of the model domain. The review shows that current models focus primarily on predicting catchment responses during a recovery period (within-burn timescales), a relatively short temporal window during which rainfall is an important source of uncertainty. At longer (between-burn) timescales, the fire regime itself, and not just fire severity, becomes a variable component of the model. At this temporal scale, the catchment processes respond to variations in the frequency and severity with which a landscape is conditioned (or ‘primed’) by fire and rain storms. Conditioning is a stochastic process that is determined by the spatial-temporal overlap of fire disturbance and rain storms. The translation of overlaps to hydro-geomorphic responses is a function of intrinsic catchment attributes (e.g. permeability, slope and catchment area). Capturing the stochastic interplay between fire and rain storms is important when land-management questions shift towards the issues of climate change and landscape-scale interventions such as prescribed burning. The review therefore includes a discussion on fire and rainfall regimes as variables which drive decadal and regional variability in hydro-geomorphic processes.

Keywords

erosion fire effects fire regime hydrology land management landscape post-fire debris flows rainfall water resources

I Introduction

Wildland fire impacts on a wide range of social, economic and environmental values (Thompson et al., 2011). Adverse outcomes can occur from first-order impacts (injury, loss of life, property and livestock) or second-order impacts such as air pollution, hydro-geomorphic events and altered ecosystem processes (Reinhardt et al., 2001). It is the probability of these adverse outcomes which determines the risk associated with the impacts of wildland fire on assets and ecosystem services (Haimes, 2009; Hyde et al., 2012). Fire is a landscape process which operates as an interaction between vegetation, climate, ignition rates, topography, and land-management strategies (Adams, 2013; Gill and Allan, 2008; Parisien and Moritz, 2009). This interaction presents many uncertainties and results in a complex decision-making environment with short- and long-term considerations across a large number of stakeholders and agencies (Hyde et al., 2012; Luce et al., 2012; Rieman et al., 2003; Thompson and Calkin, 2011; Wilson et al., 2011).

Erosion, debris flows and flash floods are hydro-geomorphic processes that operate at the intersection between sloping terrain, burned areas and intense rainfall (Cannon, 2001; Lane et al., 2006; Meyer et al., 1992; Moody and Martin, 2001a; Pierce et al., 2004; Scott and Van Wyk, 1990; Swanson, 1981; Wohl and Pearthree, 1991). The intensification of these processes after fire is a second-order effect that results from first-order effects on soil and vegetation (Hyde et al., 2012; Reinhardt et al., 2001). The effects of fire on catchments appear in time and space as perturbations which diminish at some rate during recovery (Phillips, 2009; Prosser and Williams, 1998). If there is going to be a hydrologic response, it initially starts out strong, and declines over time, with the biggest responses usually occurring in the first few years. Recovery represents a ‘window of disturbance’ or a period of reduced thresholds for large hydro-geomorphic responses (Cannon et al., 2011; Lane et al., 2006; Moody, 2012; Nyman et al., 2011; Prosser and Williams, 1998; Wondzell and King, 2003), and hence increased probability of adverse effects such as land and water degradation, disruptions to water supply systems and damage to infrastructure (Bisson et al., 2003; Cannon and Gartner, 2005; Emelko et al., 2011; Luce et al., 2012; Smith et al., 2011).

Surface roughness, hydraulic conductivity, sediment availability and slope stability are some properties (or geomorphic state variables) that can be strongly affected by fire and which contribute to increased frequency of high-magnitude events (Ebel et al., 2012; Istanbulluoglu et al., 2004; Moody et al., 2008; Nyman et al., 2010; Robichaud, 2005; Sheridan et al., 2007). There is a large body of research that combine soil measurements, hillslope experiments and models to link dynamic state variables to changes in hydrologic and geomorphic processes at various scales after wildland fire (see reviews: Moody et al., 2013; Shakesby and Doerr, 2006; Wondzell and King, 2003). Research efforts in this area are often associated with land-management agencies and therefore motivated by a need to better understand and manage the potential impacts from erosion, flash floods and debris flows. Data on post-fire responses and knowledge of underlying processes provides the basis for developing models to support effective risk-based decision-making (Hyde et al., 2012; Robichaud and Ashmun, 2012). However, the combination of (1) transient fire effects, (2) the uniqueness of each fire-rainfall sequence and (3) the spatial variation in the underlying landscape presents a major challenge in terms of generating and synthesizing data for model development (Moody et al., 2013).

Models of hydro-geomorphic processes are developed to reduce or incorporate uncertainties that operate at different timescales. For instance, the hydro-geomorphic processes in burned areas can be modelled for a specific period after a specific fire event, or it can be modelled over time for a fire regime. The two temporal scales comprise different processes and different sources of uncertainty, hence different modelling demands, including how space, time and processes are represented (Beven, 2000; Gill and Allan, 2008; Phillips, 2009). This issue of scale and model uncertainty is important when formulating risk problems and developing models, particularly if the focus question is concerned with detecting the effects of change (e.g. climate change or management intervention) (Beven, 2001b; Sivakumar, 2011). Models that operate at short temporal scales are usually concerned with the processes and properties that drive variability in catchment response after a fire event. This scale of modelling has a long history in the hydrological sciences and is concerned with transforming rainfall to a hydrologic or geomorphic response (Jakeman and Hornberger, 1993; Merritt et al., 2003). At larger and longer spatial-temporal scales, there is an increased emphasis on the fire regime itself, the rainfall regimes and their interaction with local catchment properties such as vegetation, soil and landform (Gill and Allan, 2008; Luce et al., 2012; Phillips, 2009). The paucity of literature around the effect of regimes (fire and rainfall) on catchment processes is surprising given the emerging concerns regarding landscape-scale effects of climate change and prescribed burning (Adams, 2013; Bradstock et al., 2008; Cary and Banks, 2000; Gill and Allan, 2008; Goode et al., 2012; Luce and Rieman, 2010; Luce et al., 2012; Moritz et al., 2012; Piñol et al., 1998; Stephens et al., 2012; Williams et al., 2001).

The aim of this paper is to review existing hydro-geomorphic response models for burned catchments and evaluate their application in different land-management contexts. Section II defines risk and provides an overview of risk as concept for linking research and land-management questions. This section then reviews a sample of hydro-geomorphic response models for burned areas, focusing in particular on modelling approach, the input requirements, the outputs and how the information is used in a land-management context. Section III explores how land-management questions and the decision-making space can affect the scale of models, and model demands in terms of process representation and structure. Section IV provides a landscape-scale perspective on wildland fire and its effects on catchment processes, outlining some challenges and recent developments with regards to modelling the spatial and temporal properties of fire and rainfall regimes.

II Hydro-geomorphic response models for burned areas

1 Defining the risk picture

Effective integration of land-management decisions and research outputs requires (1) a clear definition of focus question, (2) explicit conceptual modelling, (3) a suitable modelling strategy and (4) a formal scenario analysis approach (Liu et al., 2008). A clear definition of focus question provides a basis for developing a conceptual model in which opportunities, risks and uncertainties can be identified. This in turn determines the appropriate spatial and temporal scales, the forcing variables, and the components/processes that need to be represented as part of the risk picture (Liu et al., 2008). This section defines risk and outlines how risk as a concept can help facilitate effective utilization of research and models in land management.

Risk measures the probability of consequence (Aven, 2011). Analysis of risk helps identify initiating events (or threats) and conceptualize cause and effect in a decision-making environment. It is a framework for developing models that minimize complexity and sources of uncertainties in relation to the input and output variables that are most relevant to a particular focus question (Haimes, 2009). As such, risk is a platform for communication that helps ensure that scientific models are effective at identifying opportunities and evaluating the impacts of management intervention or forced changes to input variables (e.g. due to climate change). A clear conceptual model of risk means that research priorities and objectives are more likely to be aligned across researchers, modellers and the demands from land managers (Hyde et al., 2012; Liu et al., 2008; Thompson and Calkin, 2011).

Initiating events are undesirable events that lead to adverse consequences. In context of hydro-geomorphic processes, the initiating event can be defined in general as any catchment response which impacts on humans, infrastructure or ecosystem function. Bow and tie diagrams provide a risk picture which illustrates the links between causes, the initiating events and the consequences for a particular system at risk (Figure 1). When coupled with measures of vulnerability of assets, the magnitude and frequency of initiating events provide a basis for quantifying risk (Haimes, 2009; Nott, 2006). The risk picture helps define the focus question by identifying: (1) the processes underlying the initiating event, (2) the spatial and temporal scale and (3) the relevant management interventions. An important feature of risk is that the forcing variables, processes and management interventions contributing to risk (and opportunities) are dependent on time and space. For instance, the opportunities and risks are very different when faced with a current fire disturbance (i.e. timeframe of 1–10 years) as opposed to a fire regime (i.e. timeframe of 10–100s of years). Similarly for space, the opportunities and risks are different when faced with a single catchment (1–10 km²) as opposed to multiple catchments in a landscape (10–100s of km²).

Figure 1.

Contamination of a water supply system as an example of a focus question, presented as a bow and tie diagram which describes the risk picture. The focus question in this example is concerned with the probability that contamination at the water off-take exceeds treatment capacity. Horizontal bars represent causes and consequences; vertical bars represent barriers and controls such as the landscape, erosion control works and treatment capacity. The initiating event (or threat) at the centre of the diagram is the concentration of a contaminant in the water supply reservoir. The overall threat in this risk picture is a function of the magnitude and frequency with which the initiating event occurs. Treatment and erosion control represent opportunities.

Quantifying the overall risk associated with erosion in burned areas, as shown for a water supply systems in Figure 1, is a complicated task requiring information on:

rainfall and fire regimes operating in the contributing catchments;

expected hydro-geomorphic processes underlying the transport of contaminants from the burned area to the reservoir;

sediment dynamics within the reservoir itself;

treatment capacity once water is sourced at the off-take.

All these elements represent sources of uncertainty and are shown as horizontal bars along the top of the bow and tie diagram in Figure 1. Predicting each of these elements is generally not feasible and researchers therefore face the challenge of simplifying the system to reflect the key elements that contribute to uncertainty around a particular focus question. Simplification is about balancing modelling capabilities with data availability and land managers’ needs. These needs are typically concerned with those elements of the system where there is leverage for mitigating risk (i.e. an opportunity for management intervention).

Hydro-geomorphic response models can be categorized based on the type of initiating event that they are designed to predict (Table 1). The following sections provide an overview of some of these models that are used to evaluate risk associated with burned catchments. The aim here is not to evaluate how well the models perform in their applications. Instead, the focus is on outlining the different modelling appr-oaches and the processes that are represented. The models are described with sufficient detail to appreciate the dominant processes which contribute to the outputs. The examples are not an exhaustive list of all post-fire response models, but they cover a range of approaches which are representative of what is currently available for predicting runoff and erosion from burned areas.

Table 1.

Sample of models for predicting hydrologic responses from burned areas.

Initiating event	Model	Predictions	Controls/input	Spatial scale	Temporal scale
Soil loss Water contamination Sedimentation	Revised Universal Soil Loss Equation (RUSLE) (Fernández et al., 2010; Miller et al., 2003b; Renard et al., 1991)	Annual soil loss from hillslopes Large-scale assessment of erosion susceptibility	Topography, soil texture, vegetation, rainfall regime	Hillslope	Annual average
	Erosion Risk Management Tool (ERMiT) (Elliot et al., 2001; Miller et al., 2011; Robichaud et al., 2007)	Sediment yield from hillslopes Exceedance probability for post-fire hillslope erosion events during recovery Effect of control works	Topography, soil hydrologic properties (spatially variable), fire impacts, recovery, rainfall regime, control works	Hillslope	Daily events during recovery (0–3 years)
	Fire Enhanced Runoff and Gully Initiation Model (FERGI) (Istanbulluoglu et al., 2003)	Post-fire gully initiation and gully length Effect of control works	Topography, soil hydrologic properties (spatially uniform), rainfall regime, control works	Upland catchment <10 km²	Event-based
Debris flow	Statistical debris flow hazard models (Cannon et al., 2010a, 2010b; Gartner et al., 2008; Pak and Lee, 2008)	Probability and magnitude of debris flows Area of debris flow inundation	Soils, topography, fire severity, rainfall, design storm	Upland catchments (0.02–23 km²)	Event-based
	Physical modelling (Elliott et al., 2004; O’Brien et al., 1993)	Depth and area of debris flow inundation	Topography, channel configuration, discharge and sediment bulking coefficient	Catchment and fan	Event-based
Flooding	Analytical peak flow model (Moody, 2012; Moody et al., 2008)	Peak discharge for 30-minute post-fire rainfall event	Fire severity (dNBR), hydraulic connectivity, rainfall (I30) catchment area, design storm	Catchment (0.25–26.8 km²)	Event-based
	Curve number based evaluation (Livingston et al., 2005)	2-, 5-, 10-, 100-year flood post-fire	Post-fire curve number and recovery, design storm	Curve-number (CN) gauging station	Event-based

2 Erosion, water quality and sedimentation

a Annual hillslope erosion

The Revised Universal Soil Loss Equation (RUSLE) model (Renard et al., 1991) was designed to predict annual soil loss (Mg ha^-1y^-1) from hillslopes due to rainsplash and runoff. It was originally developed for agricultural systems but its use is now widespread across different land uses and natural environments including burned landscapes (Fernández et al., 2010; Larsen and MacDonald, 2007; Miller et al., 2003b; Myronidis et al., 2010). The model estimates annual soil loss as the product of rainfall erosivity, soil erodibility and non-dimensional factors that account for topography, land use and crop management. In absolute terms the predictions reflect an annual average response and the model is not intended to consider the magnitude of individual events. Thus, it is unsuitable for predicting event-based impacts.

RUSLE is based on parameters which are poorly suited for representing fire effects (Larsen and MacDonald, 2007). However, the model represents some basic topographic and soil information and therefore can be effective at distinguishing between areas of high and low erosion potential, despite the lack of accuracy in absolute terms. As such it can be a useful tool for prioritizing locations of and prescriptions for post-fire rehabilitation efforts (Fernández et al., 2010; Myronidis et al., 2010) and assessing the effectiveness of erosion control strategies (Miller et al., 2003b). RUSLE and other similar models are readily coupled with GIS and remotely sensed data on fire severity to produce landscape-scale assessments of yearly post-fire erosion potential (Chafer, 2008; Miller et al., 2003b; Sheridan et al., 2009; Vafeidis et al., 2007). This type of assessment can produce a conceptually representative measure of the relative erosion potential across burn areas.

b Event-based hillslope erosion

The Erosion Risk Management Tool (ERMiT) is a distributed and event-based hydrologic model which uses the Water Erosion Prediction Project (WEPP) to model runoff and erosion response for individual storms after a fire (Elliot et al., 2001; Flanagan et al., 2007; Robichaud et al., 2007). WEPP was initially developed in an agricultural context but has been advanced to better represent systems that are more variable in time and space (Elliot et al., 2001). ERMiT is specifically designed to help land managers assess the risk of damaging runoff and erosion events after fire with and without mitigation treatments such as seeding, straw mulch and erosion barriers. The model is more detailed and accurate than RUSLE in its representation of erosion processes (and fire effects), which also means that more detailed data are required for parameterization (Larsen and MacDonald, 2007).

The WEPP structure in ERMiT is modified to account for parameters which are influenced by high levels of spatial variability and the transient nature of fire effects (Elliot et al., 2001). Fire effects on soil properties are modelled as three fire severity classes (unburned, low severity and high severity) with parameters obtained from rainfall simulations and overland flow experiments (Robichaud et al., 2007, 2010). Spatial variability due to variable fire effects is introduced to the model by simulating the erosion response while varying the spatial configurations of homogenous tiles (or overland flow elements) along the hillslope (Elliot et al., 2001; Robichaud et al., 2007). Recovery is incorporated through yearly adjustments to the soil parameters and the configuration of overland flow elements (Robichaud et al., 2007).

ERMiT combines the event-based WEPP response model with storm outputs from a stochastic climate generator (CLINGEN) to simulate the full range of potential post-fire erosion outcomes for a sequence of yearly post-fire increments (Robichaud et al., 2007). The 24-hour temporal resolution of CLINGEN means that the predictions capture some of the event-based nature of post-fire erosion while incorporating the stochastic properties of rain storms. The distribution of possible erosion outcomes in the post-fire period is presented as erosion exceedance probabilities for events within individual years during the recovery. Coupling WEPP with CLINGEN has been used to estimate annual soil loss from hillslopes across large regions, providing a more explicit representation of processes than is possible with RUSLE (e.g. Miller et al., 2011).

c Event-based channel and gully erosion

The Fire-Enhanced Runoff and Gully Initiation Model (FERGI) combines a stochastic climate generator and a deterministic geomorphic model to estimate the probability of post-fire rainfall excess, peak flow and gully initiation in catchments with and without contour felled logged barriers (Istanbulluoglu et al., 2002, 2003). The model is most commonly used to predict peak flows, and is essentially a modified rational method based on field measurements of water repellency. The coupling of a response model with a stochastic rainfall component is similar to ERMiT despite the processes and scale of application being different. FERGI calculates the daily water balance for the wettable soil/ash layer overlying water-repellent soils incorporating the depth of wettable soil, infiltration rate, precipitation, evaporation, runoff and daily climate inputs. Sub-daily rainfall is sampled from daily rainfall totals to provide a series of storm intensities and durations which describes rainfall at a higher temporal resolution than the daily weather generator. Runoff is routed along homogenous rectangular hillslope units using the kinematic wave equation on sub-hourly timescales during storms. The hydrologic outputs are produced on a daily basis.

The model represents hillslope processes in headwaters in order to find the upper end of channel segments, determined by threshold-based channel initiation conditions. Outputs from FERGI include peak flow (m³ s^-1m^-1), gully length (m) and the effectiveness of log barriers in preventing gully erosion as a function of storm return periods. Log barrier treatment is represented in the model by adjustments to the water-storage potential on the hillslope and the fractional area that is water-repellent. The model does not provide direct estimates of erosion but assumes that once a gully has been initiated the sediment transport rate is at capacity (Istanbulluoglu et al., 2003).

3 Debris flows

a Statistical hazard models

Debris flows contain high concentrations of sediment resulting in rheology and flow processes that are different from stream flow and clear-water floods (Julien and Lan, 1991). Debris flows are extreme post-fire catchment responses and can impact directly on people and infrastructure (Cannon and Gartner, 2005). The degree of damage is linked to the area that is inundated or impacted by the debris flow at the catchment outlet (Cannon et al., 2010a). The large amount of sediment generated during debris flows also means that downstream impacts on water quality can be large (Smith et al., 2011). The management questions underlying debris flow processes are concerned with responses at the catchment outlet (0.02–23 km²), which means that the modelling focus shifts from hillslopes (e.g. ERMiT and RUSLE) and headwaters (e.g. FERGI) to the channels and their supporting catchments.

Cannon et al. (2010b) describe a model for predicting the magnitude and probability of debris flows from burned catchments. The debris flow magnitude, or the mean volume (m³) of material at the catchment outlet, was modelled empirically using measurements of debris flow erosion in 55 recently burned catchments in the intermountain western USA (Utah, Colorado and California) (Gartner et al., 2008). The independent variables in the model include total storm rainfall and the catchment area with (1) steep gradients (≥30%), and (2) high or moderate fire severity. Debris flow probability was determined by fitting binomial occurrence data with a logistic regression, using information on topography (slope and ruggedness), burn impact, rainfall intensity and soil properties (clay content and liquid limit). The rain intensity in the model can be linked to design storms by identifying intensity values for <1-hour storms with recurrence intervals between two and 10 years.

b Physically based debris-flow routing

A strong physical basis for modelling post-fire debris flows is lacking. The debris flows typically occur in response to surface runoff and erosion through a process of progressive sediment bulking. Slope failure models are therefore unsuitable for predicting the initiation of this type of debris flow response. However, the downstream hazard from post-fire debris flows has been modelled using Flo-2D (Elliott et al., 2004; O’Brien, 2000; O’Brien et al., 1993). Flo-2D is a two-dimensional flow routing system which can represent hyper-concentrated flows without violating assumptions relating to different flow regimes. The model requires an inflow hydrograph from a rain storm and a pre-determined bulking factor (sediment concentration) to simulate a debris flow. The sediment concentration has to be determined a priori and requires background information on the type of debris flow response likely to occur (Elliott et al., 2004). The inflow hydrograph is defined at the point where the flow is considered to shift from clear-water flow to hyper-concentrated flow. This point is defined somewhat arbitrarily since the initiation process is characterized by a range of flow conditions, steep flow fronts and pulses with different sediment concentrations (Gabet and Sternberg, 2008; Rickenmann, 1991).The model output from Flo-2D include runout length, depositional depth and the inundated area. The outputs have not yet been verified with real data from post-fire debris flows.

4 Peak flows

a Analytical peak flow model

Moody (2012) developed an analytical model for predicting peak flows from upland catchments based on paired rainfall and runoff measurements from 19 mountain basins in the western USA, representing different rainfall regimes and different geophysical settings. The model predicts the peak discharge per unit area Q_u (m³s^-1km^-2) from small upland catchments (0.25–26.8 km²) given a 30-minute rainfall intensity (I₃₀ ) and a known fire severity. The model accommodates different levels of detail in representation of fire impact. At the most basic level, the fire impact is given by the proportion of catchment burned, and represented through changes in a runoff coefficient, which determines the conversion of I₃₀ to unit discharge. At a more complex level, the model uses the normalized burn ratio (ΔNBR) to incorporate information on the spatial configuration of burn severities along a hillslope (Moody et al., 2008). The spatial pattern of fire severity along flow paths is parameterized through a hydraulic functional connectivity parameter. The hydraulic functional connectivity provides a more realistic representation of fire effects than the modified runoff coefficient which assumes a uniform and spatially averaged fire impact. In the third level of complexity the effect of recovery is represented through yearly adjustments to the peak flow parameters (runoff coefficient and functional connectivity).

b Curve number

The curve number (CN) approach has been used for predicting runoff in undisturbed catchments (Boughton, 1989; Cronshey et al., 1986). The method relies on a curve number (CN) parameter derived from data sets of paired rainfall-runoff measurements in small upland catchments. This parameter provides an estimate of how much runoff (in mm) is generated per every mm of rainfall after the catchment storage potential is depleted. Curve-number parameters have been generalized for catchments based on their soil hydraulic properties, vegetation and hydrologic condition. The depth of runoff is converted to a measure of peak discharge by incorporating a time of concentration parameter t_c which is influenced by surface roughness, channel configuration and topography.

Fire effects on peak flows can be quantified by estimating CN for pre- and post-fire conditions (Cerrelli, 2005; Elliott et al., 2004; Livingston et al., 2005; Springer and Hawkins, 2005). The degree of change in CN between pre- and post-fire condition is varied to reflect different fire severities and different sensitivities to fire impacts. The t_c parameter is kept constant so the effect of fire on flow velocity is not represented in the method (Cerrelli, 2005). Recovery can be incorporated by making CN return gradually towards pre-fire levels as a function of time (Livingston et al., 2005). Finally, the model outputs can be combined with design storms to produce post-fire peak flow predictions in probabilistic terms (i.e. 5-, 10-, 100-year flood).

The post-fire CN parameters are derived at a gauging station as a function of a single and assumed homogenous fire severity. Within catchments, however, the fire can have a range of impacts depending on both fire extent and severity. Spatially variable fire effect is taken into account by adjusting the estimated peak flow according to the proportion of area burned (Cerrelli, 2005). This approach, however, is not substantiated with data and relies on judgement and the best information available. No explicit representation of catchment properties and the lack of consistency when determining post-fire curve number means that the approach is constrained in its capacity to be generalized beyond the exact catchment conditions and scales where parameters were obtained (Moody, 2012; Springer and Hawkins, 2005).

III Resolution, scale and uncertainty in models

Resolution and scale determine how fire effects, rain storms and hydrologic processes are represented in models (Figure 2a). Resolution is the level of detail in process representation. Scale is the temporal and spatial extent of the modelling space. Variations in resolution and scale are driven by (1) purpose of modelling (or focus question), (2) the data availability and (3) the trade-offs between predictive power and explanatory depth (Beven, 2001a; Dawdy, 2007; Sivakumar, 2008). Each model listed in Table 1 has inputs, outputs and model structures which are aligned with specific management objectives, so that processes are represented at a scale and resolution which is appropriate for evaluating risk and opportunities relating to a particular focus question. This section examines how focus questions and scale affect resolution and uncertainty in models. The discussion is structured around scale, beginning with models that describe fire effects on plot- and hillslope-scale processes, and concluding with catchment-scale models that are concerned with responses at timescales when the fire regime itself is a variable component of the modelling space.

Figure 2.

(a) Dominant factors that affect fire-related erosion at different spatial and temporal scales. The slope, topography and landscape factors represent the land surface and all the associated properties which influence the sensitivity of the system to fire effects and which determine the conversion of rainfall to runoff and erosion. The figure has been adapted from the fire regime literature where there has been a strong research focus on the role of scale in determining shifts in dominant fire processes (e.g. Parisien and Moritz, 2009). (b) The spatial and temporal scales of hydro-geomorphic response models described in the literature.

Within hillslopes at a plot scale it is the soil physical properties and the instantaneous rainfall intensity that determine the response (Figure 2a). A very high spatial-temporal resolution is therefore required to represent internal parameters and processes (Moody and Ebel, 2013) (Figure 2b). Models at this scale serve as tools for understanding processes, thus providing a stronger physical basis on which to develop larger-scale response models (Table 2). The high spatial-temporal resolution is sometimes carried through from plot scale to larger scales because of the need to explicitly capture and represent the hydrologic effects of management interventions such as log barriers and mulching (e.g. FERGI and ERMiT; Istanbulluoglu et al., 2003; Robichaud et al., 2007) (Figure 2b). High resolution in process representation at larger scales is made possible through distributed models that represent the transfer of water and sediment across model elements. With increasing scale, however, there is extra complexity due to spatial burn patterns (area burned at different severities), storm properties (size, durations and intensity) and the catchment configuration (slope, relief, channel configuration and variation in soil properties) (Kean et al., 2011; Luce, 2005; Moody and Martin, 2001b). This means increased data demands and increased uncertainty around the internal processes leading to a response (Michaud and Sorooshian, 1994). In cases where data and process understanding is limited, or where high resolution is unnecessary, the alternative approach is to lump models and calibrate the catchment response directly with parameters describing the contributing area and storm properties as a whole (Cannon and Gartner, 2005; Cannon et al., 2010b; Elliott et al., 2004; Livingston et al., 2005; Moody, 2012). Lumped models reflect a priority towards prediction of catchment-scale response, as opposed to understanding and representing the internal processes contributing to the catchment response (Dawdy, 2007) (Table 2). The level of detail in the parameterization is sufficient for predicting the fire-impact and catchment-scale responses which emerge implicitly as a result of internal (or intra-catchment) processes. Focus questions in these types of models are typically concerned with engineering design, warning systems and evacuation procedures.

Table 2.

Examples of focus questions at different combinations of spatial and temporal scale.

Spatial scale	Temporal scale	Focus question
Plot	Minutes	How does fire intensity affect runoff and erosion? How to represent key processes underlying catchment response? What is the relation between soil heating and fire effects on hydro-geomorphic processes?
Hillslope and catchment	Minutes	What is the magnitude of response from a burned catchment for a given storm? Does a burned catchment represent risk to assets or ecosystem values? What is the effect of management intervention on the response of a catchment?
Hillslope and catchment	Years	What is the probability of a catchment producing an undesirable event during recovery? When and where is the risk highest? What are the opportunities in terms of management interventions following a fire?
Hillslope and catchment	Decades	How do catchment processes respond to changes in fire and rainfall regimes? What is the long-term effect of fire on sediment delivery from upland catchments? How does fire disturbance compare to other forest disturbance effects? How should post-fire responses be considered in design of infrastructure?
Region	Decades	How does the application of prescribed fire affect catchment processes over time? What is the frequency and magnitude of fire-related responses in a particular river reach? How does frequency and magnitude of sediment delivery from burned areas compare to sediment delivery from other land uses?

Focus questions sometimes demand models which predict catchment processes over timescales that extend beyond the hydro-geomorphic processes underlying each individual event. This introduces uncertainty around how often and how strongly a catchment is affected by fire and rain storms. The modelling task therefore shifts from magnitude (‘how much’) to magnitude and frequency (‘how much and how often’) (Benda and Dunne, 1997; Cannon et al., 2010b; Istanbulluoglu et al., 2004; Jones et al., 2011; Robichaud et al., 2007). Predicting longer-term processes requires model structures that capture between-event uncertainty as opposed to the within-event uncertainty underlying each individual event. Uncertainty within events is reduced in response models that represent the internal hydrologic properties that determine transfer processes and event magnitude (Figure 3, a and b). This uncertainty is epistemic (i.e. can be reduced with more data and better response models) and stems from the incomplete understanding and lack of data on the processes that constitute the catchment response. Between-event uncertainties are concerned with the frequency and severity with which the system is ‘primed’ (or conditioned) by fire and rain storms (Figure 3c). This process is driven by stochastic processes (i.e. the uncertainty is aleatoric) and must be incorporated into the model structure and coupled with the hydro-geomorphic response models.

Figure 3.

Sources of uncertainty when modelling hydro-geomorphic responses from burned areas. Uncertainties are introduced through: (a) topographic (or landform) variability, (b) spatial variability in surface properties and (c) temporal variability at various timescales of modelling.

Stochastic rainfall generators have been coupled with hydro-geomorphic response models to produce post-fire erosion predictions that incorporate the between-event uncertainty associated with a particular rainfall regime. In ERMiT and FERGI, for instance (see Table 1), the effect of stochastic rainfall is captured using long-term rainfall statistics, and incorporated into the simulations of runoff and erosion. The temporal uncertainty in hydro-geomorphic responses in a particular catchment is expressed in terms of the exceedance probability during recovery from the fire disturbance. The rainfall events are generated using data from local weather stations and therefore capture the effect of spatial variability in rainfall regimes. Stochastic uncertainty in rain storms during recovery can also be represented by coupling event-based predictions with design storms (Cannon et al., 2010a). The probability of exceedance during recovery is sensitive to the timescales over which the system is modelled because the fire impact decreases with time (Figure 3c).

The initial fire impact in post-fire response models is a given, so the temporal window is restricted to the timescale of recovery (usually <5 years but sometimes up to 10 years or longer) (Shakesby and Doerr, 2006; Swanson, 1981; Wondzell and King, 2003). Efforts to model response at this timescale reflect a demand for information on expected responses from hillslopes and catchments after a fire has occurred or with an assumed (or given) fire severity distribution. This scale of investigation represents a relatively narrow spatial and temporal perspective in context of fire regimes and its interaction with hydrologic and geomorphic processes. The implication (in terms of process representation) of such a limited temporal perspective is that the full landscape-scale interaction between fire impacts, catchments and rainfall is poorly understood. Models that operate during the ‘window of disturbance’ are suitable for quantifying the response to varying levels of fire impact (a disturbance), although fail to capture the effects of shifts in the fire regime itself (a persistent change) (Phillips, 2009).

When models treat the fire regime as a variable component, then the catchment processes are determined by rainfall regimes in combination with the fire regime and landscape characteristics such as tectonics, geology and vegetation (Moody and Martin, 2009; Swanson, 1981). At this longer temporal scale, factors such as climate change, fire management and changing patterns of land use will influence the disturbance regime, and hence the expected hydro-geomorphic processes operating in a catchment (Bradstock et al., 2009; Cary et al., 2009; Hennessy et al., 2005; Luce et al., 2012). For instance, the reduced rainfall and increased frequency of high temperatures due to climate change can lead to more days with extreme fire conditions, thus resulting in increased fire frequency in places such as southeast Australia (Hennessy et al., 2005), the Mediterranean (Mouillot et al., 2002) and parts of the western USA (Westerling et al., 2006). At the same time, the widespread application of prescribed fire can increase in the area that is burned at any one time but reduce the average annual area burned by wildfire as a result, due to a trade-off between frequent, low-intensity fire and infrequent, high-intensity fire (Bradstock and Williams, 2009; Bradstock et al., 2012; Price et al., 2012). The rainfall regime may also undergo changes with climate change, with some fire-prone regions expected to see increased frequency of high-intensity rain storms (Alexander and Arblaster, 2009; Hennessy et al., 1997).

Focus questions at this broad spatial-temporal scale are concerned with: (1) how climate change translates to changes in hydro-geomorphic processes and probability of adverse impacts; (2) how these changes interact with increased regional-scale application of prescribed fire; and (3) what this means for responses within a single catchment versus that of large basins with multiple contributing subcatchments. The process representation for these types of questions must be aligned to correspond with:

size and frequency of storm cells;

size and frequency of fires;

variation in surface properties and landforms.

These are characterized at scales that are much larger than the internal factors which drive catchment or hillslope processes within events. When compared to plots- and hillslope-scale studies of fire effects and post-fire processes, there have been relatively few studies that explore how the effects of fire and rainfall regimes can be measured and represented in models. This is somewhat surprising given that land management and climate change are having a large impact on the way in which fire operates as a disturbance agent (Adams, 2013; Bradstock et al., 2008; Cary and Banks, 2000; Gill and Allan, 2008; Goode et al., 2012; Luce and Rieman, 2010; Luce et al., 2012; Moritz et al., 2012; Piñol et al., 1998; Stephens et al., 2012; Williams et al., 2001). The next section therefore explores the dominant patterns and processes that emerge when fire effects are viewed from a landscape-scale perspective.

IV Hydro-geomorphic responses under variable fire and rainfall regimes

1 Fire, rain storms and ‘episodic patches of activity’

‘Episodic patches of activity’ is a pattern which describes the hydro-geomorphic processes that emerge when fire impacts and rain storms are viewed at a landscape scale (Miller et al., 2003a). In this description by Miller et al. (2003a), the ‘activity’ is concerned with uncertainties underlying the hydrologic and geomorphic transfer processes in a particular catchment. ‘Episodicity’ is a source of temporal uncertainty, and a property which results from the stochastic interplay between burned areas and storms. ‘Patchiness’ in response is driven by the spatial intersection between storms and burned areas and the subsequent interactions with surface properties that control runoff generation and sediment availability. This section revisits the idea that fire results in ‘episodic patches of activity’, and examines how rain storms and fire can be modelled in space and time to capture the processes leading to this type of pattern.

The activity is a function of the variables that control transfer processes and that determine catchment response to rainfall. These variables typically include topographic properties (e.g. slope, ruggedness), soil properties, the fire impact, and the hyetograph from the rain storm. The response can be modelled at various scales through hydrologic transfer processes and distributed models or through empirical relationships between rainfall, catchment properties and a response (Istanbulluoglu et al., 2004; Larsen and MacDonald, 2007; Robichaud et al., 2007). The models are designed to reduce the uncertainty in how rainfall translates to a response in systems with known fire impacts. The fire impact is a departure of state variables from stable background conditions and depends on (1) the severity of the fire, (2) the sensitivity of the system to fire severity and (3) the state of recovery (Prosser and Williams, 1998).

Episodicity is about the temporal pattern of responses which emerge as a result of fire and rainfall regimes (Kirchner et al., 2001; Meyer et al., 2001; Pierce et al., 2004; Tomkins et al., 2007) (Figure 3c). Modelling studies that aim to represent the net effect of fire as a disturbance agent over time therefore require long simulations or analytical solutions in order to capture the full range (or long-term average) of potential post-fire outcomes (Benda and Dunne, 1997; Istanbulluoglu et al., 2004; Jones et al., 2011; Luce et al., 2012; Smith et al., 2009). Failing to capture the episodic nature of catchment responses can lead to misrepresentation of disturbance effects in catchment processes, particularly in settings where the geomorphology is dominated by large and infrequent events (Istanbulluoglu et al., 2004; Kirchner et al., 2001; Pierce et al., 2004).

Patchiness is driven by the spatial correlation in catchment properties and spatial structures in fire severity and rain storms. Two adjacent catchments within any burn, for instance, are likely to have similar properties and receive comparable (spatially correlated) fire and rainfall conditions, and are therefore more similar in their processes than catchments separated by long distances. This spatial dependence and subsequent patchy response means that the magnitude and frequency of events within each individual catchment is not representative of the total or net effect of burned areas on the landscape response (Luce and Rieman, 2010). Across a large landscape, the spatial scaling of fire disturbances and interacting rain storms means that some nature of disturbance is generally pretty common within a large area (e.g. within a large watershed), but that disturbances among areas tend to be somewhat asynchronous. This is because the response within each catchment is embedded within the scale of storm cells and burn areas, both of which operate at larger spatial scales than the catchment response itself.

Patchiness and episodicity are dependent on the stochastic interplay between storm cells and burned areas. This spatial-temporal process is a function of size, location and frequency of storms and fires. These parameters define the rainfall and fire regime for an area. Regime parameters have in the past mainly been characterized through the statistical properties of long-term data records at a point (Johnson and Gutsell, 1994; Svensson and Jones, 2010), making a full spatial-temporal representation of fire and storms impossible. However, new measurement techniques and increasing modelling capacity are improving the way in which fire and storms can be represented in space and time. This has large implications for how fire effects on catchment processes can be modelled at a landscape scale where the response over time is episodic and patchy (e.g. Jones et al., 2011). The following sections therefore provide an overview of recent developments in the modelling of fires and storms as spatial-temporal processes.

2 Representing fire in space and time

Essentially all fire effects on a hydrologic system can be traced back to the combustion of duff, litter and vegetation and the associated heat impacts on the soil system (Hungerford et al., 1991; Neary et al., 2005). Impacts on catchments due to fire can be measured indirectly using fire intensity (energy output during the fire) or fire severity (loss of organic matter) as fire metrics (Chafer, 2008; Keeley, 2009; Key and Benson, 2005; Parsons et al., 2010; Wooster et al., 2005). Fire intensity can be characterized at large scales by estimating fire energy release from measurements of middle infrared and thermal infrared radiation using air- or space-borne radiometers (Dennison et al., 2006; Wooster et al., 2005). Fire severity can be measured from vegetation indices calculated at large scales using aerial- or satellite-based imagery. Metrics of intensity and severity are not directly linked to hydrology and must be calibrated with measured impacts on hydrologic properties, processes or a response variable (Chafer, 2008; Parsons et al., 2010). The quantification of fire impact is therefore specific to a fire metric and the scale at which the associated hydrologic impact is measured. The impact at a point due to soil heating may be a change in particle size, porosity and water repellency status (Doerr et al., 2004). At plot to catchment scales the impact reflects a spatial distribution of fire effects and may be characterized in terms of runoff generation or peak discharge, thus incorporating information on fire-induced changes to a range of soil properties that underlie the hydrologic response (Moody, 2012; Robichaud et al., 2007). The appropriate scale of metrics of fire impact is determined by the spatial and temporal scale at which hydrologic processes are being modelled and measured. An important consideration is whether the fire metric is used to adjust soil parameters in distributed models or if it is used directly as an explanatory variable in lumped response models.

Using pre- and post-fire spectral signals in satellite imagery (e.g. Landsat or SPOT) to map fire extent and severity is now a common procedure following wildfires (Chafer, 2008; Key and Benson, 2005; Moody et al., 2008; Parsons et al., 2010; Tanaka et al., 1983; Victorian Department of Sustainability and Environment, 2009). For low-intensity fires however, the fire severity is often patchy at small scales due to variability in fuel moisture, fuel loads and microclimate during the passage of the fire (Cawson et al., 2012; Gould et al., 2007). This variability is not captured in satellite-derived severity measurement and therefore difficult to measure directly for large areas. For high-intensity fires the heterogeneity occurs at larger scales due to factors such as topography, local weather conditions during burning and systematic variability in vegetation properties (Bradstock et al., 2010). This type of variability is captured in fire severity maps and can be used to scale fire effects according to the distribution or heterogeneity of severity within the catchment (Cannon et al., 2010b; Hyde et al., 2007; Moody, 2012). During crown fires and extreme wildfire conditions the intensity can be very high throughout the burn, thus resulting in homogenous severity (Dillon et al., 2011; Turner and Romme, 1994).

If the focus question is concerned with the hydrologic responses that may occur from burned catchments in the future, then the exact properties of future fires is uncertain. The fire severity or intensity is not given and can obviously not be mapped and characterized for a specific catchment or hillslope of interest. The modelling focus therefore shifts from characterizing the impact from specific (known) fire events towards characterizing impacts from a fire regime, where the exact nature of any given burn is unknown (e.g. Dillon et al., 2011). In a hydrologic context is it is the frequency of fire, the area of fires and the fire properties within those areas which determine the overall disturbance regime (Luce et al., 2012; Miller et al., 2003a). These regimes operate at large spatial and temporal scales where vegetation, climate and sources of ignition are the dominant drivers of fire size and frequency (Malamud et al., 2005; Parisien and Moritz, 2009).

Heterogeneities of impact within fires (i.e. fire severity) is linked to fire behaviour and processes that occur over finer spatial and temporal scales than the fire regime itself (Catchpole, 2002; Dillon et al., 2011; Parisien and Moritz, 2009). This means that the regional drivers of fire regimes and the local drivers of fire behaviour both operate within their respective spatial and temporal scales to determine the pattern of burning (severity, extent and frequency) for a particular landscape. The spatial scale at which regimes are characterized is determined by the geographical extent of distinct wildfire-prone eco-regions. This can be problematic when steep climatic gradients results in large variability in fire regimes over short distances (Clark et al., 2002). Some of this complexity can be incorporated in predictions by using data from past fires to reconstruct fire regimes not only in terms of fire frequency and size, but also the internal (or downscaled) variability in the severity (e.g. Bradstock et al., 2010; Dillon et al., 2011).

Statistical analysis of contemporary fire records shows that the relationship between fire size and fire frequency roughly exhibits a power-law relation (Cui and Perera, 2008; Malamud et al., 1998; Moritz et al., 2005). This means that that the cumulative frequency of a fire greater than or equal to size A_f can be expressed as N_cf = cA_f ^-b _. This relationship has been found to hold for fire regimes across ecosystems and geographic locations (Fiorucci et al., 2008; Malamud et al., 2005; Ricotta et al., 1999). Examples of power-law behaviour in fire regimes are given in Figure 4 for the Australian Capital Territory and Victoria in southeast Australia. Fire regimes have been characterized for large areas through regionally derived power-law relations obtained from fire history data (Fiorucci et al., 2008; Malamud et al., 2005). These type of fire regime analyses provide parameters which can be used to predict the frequency and size of future wildfires (Podur et al., 2010). However, such statistical regime models do not represent heterogeneity within burns.

Figure 4.

Cumulative frequency-area wildfire statistics for Victoria (VIC) and Australian Capital Territory (ACT) in southeast Australia for all fires with area >1 km². The frequency of fires was normalized by: (1) the total forested area of the two wildfire regions; and (2) period of observation, giving the units of fires year^-1 km^-2. When the data are presented as cumulative frequency, the fitted lines over predict the frequency of large fire events for both regions. Data for forested areas in ACT (∼1600 km²) were obtained from fire history records from 1926 to 1999 (www.firebreak.com.au/firehistory.html), including 74 fires ranging in size from 1.00 to 360 km². The wildfire history for Victoria (∼78,000 km²) was obtained from the Victorian Department of Sustainability and Environment, for the years 1972–2009, comprising 1040 fires ranging in size from 1 to 10,900 km².

The processes that constitute a fire regime are stochastic and interact over large spatial and temporal scales. The difficulty of obtaining adequate data at these scales often means that empirical analysis is unable to capture the whole spectrum of potential fire outcomes under past, current and future scenarios. Therefore, simulation models are often used to quantify the potential impacts of climate change, weather and fuel management on fire regimes (Bradstock et al., 2006; Cary et al., 2009; Finney et al., 2007; Keane et al., 2004; King et al., 2006; Li et al., 2005; Peterson et al., 2011). The key advantages of using simulation models to characterize fire regimes is that they (1) usually allow for greater temporal depth than empirical and statistical methods, (2) represent fire behaviour processes and can incorporate explicitly the effects of fuel management, fire suppression and changing climate, and (3) have the capacity to deliver tailored outputs for specific impacts (Keane et al., 2003). These features make simulation models an attractive option for managers to explore the effects of management interventions and climate change on fire disturbance.

Land-management interventions influence fire regimes primarily through fire suppression and prescribed burning. The objective of prescribed burning is to reduce wildfire hazard and the risk to communities, infrastructure and the commercial and ecological values that forest and catchments provide (Wade and Lunsford, 1989). The underlying assumption is that reduced fuel loads and reduced fuel contiguity promotes smaller, less intense and more manageable wildfires (e.g. Boer et al., 2009; Finney et al., 2007; King et al., 2006; Piñol et al., 2005; Price and Bradstock, 2011; Syphard et al., 2011). In simulation models, the variations in fire size and intensity under different climate and management scenarios are captured through the explicit representation of local fuel build-up and weather on fire behaviour in a mechanistic fire spread component (Cary et al., 2006; Massman et al., 2010).

The links between modelled fire intensity, measurable fire severity metrics, and impacts on hydrology are poorly tested and yet to be quantified (Massman et al., 2010; Reinhardt and Dickinson, 2010). However, a broad distinction between fire regimes is important when assessing the relative impacts of high-frequency, low-intensity fire regime as opposed to low-frequency, high-intensity fire regime on catchment processes. This fire regime trade-off could be explored in terms of catchment processes, using the concept of efficacy, in a similar manner to how carbon emissions were quantified for different prescribed fire scenarios (Bradstock and Williams, 2009; Vilén and Fernandes, 2011). The efficacy is the relationship between the rate of prescribed fire treatment (e.g average % of landscape treated by low-intensity fire) and the corresponding response of unplanned fire activity (e.g mean incidence, area burned and consequent rate of burning by higher-intensity unplanned fire). Efficacy parameters have been derived for fire-prone ecosystems in southeast Australia and the western USA using a combination of fire history data and outputs from simulation models (Bradstock et al., 2012; Price et al., 2012).

3 Representing rain storms in space and time

The frequency and magnitude of rain storms in small catchments can be modelled through stochastic rainfall generators, design storms or intensity frequency duration curves, all of which are parameterized from point-based rainfall records. In these approaches the storms are modelled as a single and homogenous ‘cell’ which is partly or fully embedded within a defined catchment. For large catchments, or for multiple catchments within a landscape, the erosion response following fire is likely to occur as a result of different rain storms which operate independently of each other, or as dependent clusters which are linked to a frontal system. The effect of clustering and spatial dependence in rainfall has not yet been explored in post-fire response models and is not captured in the concept of design storms. Yet for any large area it is both the spatial and temporal coincidence of fire-impacted areas and storms which determine frequency and strength with which hydro-geomorphic processes are affected by fire disturbance.

Space-time Poisson models represent rainfall as heterogeneous fields in a similar manner to how rain storms operate in the landscape (Cowpertwait et al., 2002; Cox and Isham, 1988; Rodriguez-Iturbe et al., 1986). These models are based on the ‘time-space correlation structure of rainfall’ and reflect ‘hierarchical features observed in actual rain systems, such as cells, cell clusters and rain bands’ (Sivapalan and Blöschl, 1998). Hence, these approaches may prove effective at overcoming some of the limitations with point-based rainfall models such as design storms. Space-time Poisson models can be applied as analytical tools or as continuous space-time simulation of rainfall for distributed hydrologic modelling under current and future climate scenarios (Burton et al., 2010a, 2010b).

The level of complexity required from these models depends on the application. If continuous hydrologic modelling is the objective, then it might be important to develop simulations with realistic spatial and temporal structures for all rainfall events (Chandler et al., 2006). However, if the effort is aimed at modelling the frequency and magnitude of storms across a landscape on an annual basis, then the modelling is focused on accurately representing extremes (Jones et al., 2011). The spatial variability displayed across the region remains an important component, but seasonal variability and spatial structure within storms may be less important. This simplifies the modelling task, but complicates the parameterization due to the large data requirements and problem of reproducing extremes (Cowpertwait et al., 2002).

Large data requirements are a limitation when characterizing rainfall, particularly when rain-gauge networks are sparse and orographic effects are strong, which is typically the case for mountainous regions. Also, the high-intensity storms that drive post-fire response occur at sub-hourly timescales and can be highly localized (Kean et al., 2011; Moody and Martin, 2009). Radar data provide an important source of spatially representative rainfall data at fine temporal scales. Despite the uncertainties associated with radar data itself (reflectivity measurements) and the calibration procedure (Krajewski et al., 2010; Seed et al., 2002), it has been used effectively (1) to obtain indirect measurements of actual rainfall, (2) to drive spatially distributed hydrologic models, (3) to parameterize spatial-temporal rainfall models and (4) to support real-time forecasting (Chandler et al., 2006; Jorgensen et al., 2011; Nyman et al., 2011; Rozalis et al., 2010). Combining spatial-temporal rainfall models with spatial information from radar data and temporal records from rain-gauge networks will lead to improved representation of rainfall regimes.

V Conclusions

Climate change and widespread application of prescribed fire can have large implications for how fire regimes operate as a disturbance agent in catchments. Land-management questions are therefore increasingly concerned with the role of the fire regime as a control on the frequency and magnitude with which catchments produce hydro-geomorphic events that are detrimental to water resources or infrastructure. Current post-fire response models, however, are largely designed to predict catchment processes after a fire event has occurred. The aim of these models has been to reduce deterministic uncertainties in catchment-specific response models while incorporating rainfall as a stochastic variable during a recovery window. New modelling approaches are needed in order to address the larger-scale implications of climate change and fire management on longer-term catchment processes.

A regional perspective where the fire regime is variable is particularly pertinent under current policy environment in SE Australia, for instance, where large resources are allocated towards increasing the annual prescribed burning and where climate change is likely to promote wildfire activity. A regional and longer term modelling approach means that patchiness and episodicity in response become important properties of the hydro-geomorphic system. Capturing these properties requires a shift in modelling focus to a scale and resolution which are aligned with the key factors and processes that drive regional and longer-term catchment dynamics. The frequency and magnitude of fire events, the rain storms and their interaction with variable landscapes become more important as a source of uncertainty than the exact hydrologic transfer processes that occur within a recovering catchment.

Key areas for future research and modelling of landscape-scale interactions between fires, catchment processes under changing climate and fire regimes should include:

increased emphasis on the first-order interaction between rain storms and fire events as controls on catchment responses over time and thus the forcing variables underlying risk;

the development of new tools for measuring and modelling the coincidence of fire impacts and rain storms in space and time;

linking frequency and magnitude models with measurements of longer-term erosion rates and event-frequency in fire-prone ecosystems.

Fire regimes, rainfall regimes and landscapes vary from region to region. Developing a basis for regional comparison of catchment processes in different fire-prone systems can therefore provide useful insights into the relative importance of landscapes, fire and rainfall regimes as controls on how fire interacts over time with hydrologic and geomorphic processes. Isolating the regional differences and their effects on processes means that catchments can be compared and evaluated in terms of how fire and storm alone may act as agents (or primers) of hydrologic and geomorphic events. A large-scale perspective can help reduce some of the complexity which drives fine-scale variability while allowing the key processes that drive landscape-scale differences in post-fire response patterns to emerge. Simultaneous modelling of fire and rain storms as landscape-scale processes is challenging, but the task seems increasingly feasible with new sources of data and new modelling tools for spatial-temporal representation of fire and rainfall.

Footnotes

Funding

Funded by the Bushfire Cooperative Research Centre (CRC) with additional support from the Victorian Department of Environment and Primary Industries (DEPI) and Melbourne Water.

Acknowledgements

The authors are grateful for the insightful and thorough comments from three anonymous reviewers.

References

Adams

(2013) Mega-fires, tipping points and ecosystem services: Managing forests and woodlands in an uncertain future. Forest Ecology and Management 294: 250–261.

Alexander

Arblaster

(2009) Assessing trends in observed and modelled climate extremes over Australia in relation to future projections. International Journal of Climatology 29: 417–435.

Aven

(2011) On some recent definitions and analysis frameworks for risk, vulnerability, and resilience. Risk Analysis 31: 515–522.

Benda

Dunne

(1997) Stochastic forcing of sediment supply to channel networks from landsliding and debris flow. Water Resources Research 33: 2849–2863.

Beven

(2000) Uniqueness of place and process representations in hydrological modelling. Hydrology and Earth System Sciences 4: 203–213.

Beven

(2001a) On explanatory depth and predictive power. Hydrological Processes 15: 3069–3072.

Beven

(2001b) On fire and rain (or predicting the effects of change). Hydrological Processes 15: 1397–1399.

Bisson

Rieman

Luce

. (2003) Fire and aquatic ecosystems of the western USA: Current knowledge and key questions. Forest Ecology and Management 178: 213–229.

Boer

Sadler

Wittkuhn

. (2009) Long-term impacts of prescribed burning on regional extent and incidence of wildfires: Evidence from 50 years of active fire management in SW Australian forests. Forest Ecology and Management 259: 132–142.

10.

Boughton

(1989) A review of the USDA SCS curve number method. Soil Research 27: 511–523.

11.

Bradstock

Williams

(2009) Can Australian fire regimes be managed for carbon benefits? New Phytologist 183: 931–934.

12.

Bradstock

Bedward

Cohn

(2006) The modelled effects of differing fire management strategies on the conifer Callitris verrucosa within semi-arid mallee vegetation in Australia. Journal of Applied Ecology 43: 281–292.

13.

Bradstock

Cary

Davies

. (2012) Wildfires, fuel treatment and risk mitigation in Australian eucalypt forests: Insights from landscape-scale simulation. Journal of Environmental Management 105: 66–75.

14.

Bradstock

Cohn

Gill

. (2009) Prediction of the probability of large fires in the Sydney region of south-eastern Australia using fire weather. International Journal of Wildland Fire 18: 932–943.

15.

Bradstock

Davies

Price

, et al. (2008) Effects of climate change on bushfire threats to biodiversity, ecosystem processes and people in the Sydney region. Report to the New South Wales Department of Environment and Climate Change. Climate Change Impacts and Adaptation Research Project 050831, New South Wales, Australia.

16.

Bradstock

Hammill

Collins

. (2010) Effects of weather, fuel and terrain on fire severity in topographically diverse landscapes of south-eastern Australia. Landscape Ecology 25: 607–619.

17.

Burton

Fowler

Blenkinsop

. (2010a) Downscaling transient climate change using a Neyman-Scott Rectangular Pulses stochastic rainfall model. Journal of Hydrology 381: 18–32.

18.

Burton

Fowler

Kilsby

, et al. (2010b) A stochastic model for the spatial-temporal simulation of nonhomogeneous rainfall occurrence and amounts. Water Resources Research 46: W11501.

19.

Cannon

(2001) Debris-flow generation from recently burned watersheds. Environmental and Engineering Geoscience 7: 321–341.

20.

Cannon

Gartner

(2005) Wildfire-related debris flow from a hazards perspective. In: Jakob

Hungr

(eds) Debris-flow Hazards and Related Phenomena. Berlin: Springer, 363–385.

21.

Cannon

Boldt

Laber

. (2011) Rainfall intensity-duration thresholds for postfire debris-flow emergency-response planning. Natural Hazards 59: 209–236.

22.

Cannon

Gartner

Rupert

. (2010a) Emergency assessment of postfire debris-flow hazards for the 2009 Station fire, San Gabriel Mountains, southern California. Open-File Report 2009-1227. Reston, VA: US Geological Survey, 27 pp.

23.

Cannon

Gartner

Rupert

. (2010b) Predicting the probability and volume of postwildfire debris flows in the intermountain western United States. Geological Society of America Bulletin 122: 127–144.

24.

Cary

Banks

(2000) Fire regime sensitivity to global climate change: An Australian perspective. In: Innes

Beniston

Verstraete

(eds) Biomass Burning and its Inter-Relationships with the Climate System. Amsterdam: Springer, 233–246.

25.

Cary

Flannigan

Keane

. (2009) Relative importance of fuel management, ignition management and weather for area burned: Evidence from five landscape-fire-succession models. International Journal of Wildland Fire 18: 147–156.

26.

Cary

Keane

Gardner

. (2006) Comparison of the sensitivity of landscape-fire-succession models to variation in terrain, fuel pattern, climate and weather. Landscape Ecology 21: 121–137.

27.

Catchpole

(2002) Fire properties and burn patterns in heterogenous landscapes. In: Braddock

Willams

Gill

(eds) Flammable Australia: The Fire Regimes and Biodiversity of a Continent. Cambridge: Cambridge University Press, 49–75.

28.

Cawson

Sheridan

Smith

. (2012) Surface runoff and erosion after prescribed burning and the effect of different fire regimes in forests and shrublands: A review. International Journal of Wildland Fire 21: 857–872.

29.

Cerrelli

(2005) FIRE HYDRO: A simplified method for predicting peak discharges to assist in the design of flood protection measures for Western wildfiresIn: Moglen

(ed.) Managing Watersheds for Human and Natural Impacts. Alexandria, VA: American Society of Civil Engineers, 935–941.

30.

Chafer

(2008) A comparison of fire severity measures: An Australian example and implications for predicting major areas of soil erosion. CATENA 74: 235–245.

31.

Chandler

Isham

Bellone

. (2006) Space-time modelling of rainfall for continuous simulation. In: Finkenstadt

Held

Isham

(eds) Statistical Methods for Spatio-temporal Systems. Boca Raton, FL: Chapman and Hall/CRC, 177–215.

32.

Clark

Gill

Kershaw

(2002) Spatial variability in fire regimes: Its effects on recent and past vegetation. In: Bradstock

Williams

Gill

(eds) Flammable Australia: The Fire Regimes and Biodiverstiy of a Continent. Cambridge: Cambridge University Press, 125–141.

33.

Cowpertwait

PSP

Kilsby

O’Connell

(2002) A space-time Neyman-Scott model of rainfall: Empirical analysis of extremes. Water Resources Research 38(8): 6-1–6-14..

34.

Cox

Isham

(1988) A simple spatial-temporal model of rainfall. Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences 415: 317–328.

35.

Cronshey

McCuen

Miller

. (1986) Urban hydrology for small watersheds. General Technical Report 210-VI-TR-55. Washington, DC: US Department of Agriculture, Soil Conservation Service, 164 pp.

36.

Cui

Perera

(2008) What do we know about forest fire size distribution, and why is this knowledge useful for forest management? International Journal of Wildland Fire 17: 234–244.

37.

Dawdy

(2007) Prediction versus understanding (The 2006 Ven Te Chow Lecture). Journal of Hydrologic Engineering 12: 1–3.

38.

Dennison

Charoensiri

Roberts

. (2006) Wildfire temperature and land cover modeling using hyperspectral data. Remote Sensing of Environment 100: 212–222.

39.

Dillon

Holden

Morgan

, et al. (2011) Both topography and climate affected forest and woodland burn severity in two regions of the western US, 1984 to 2006. Ecosphere 2: art130.

40.

Doerr

Blake

Shakesby

. (2004) Heating effects on water repellency in Australian eucalypt forest soils and their value in estimating wildfire soil temperatures. International Journal of Wildland Fire 13: 157–163.

41.

Ebel

Moody

Martin

(2012) Hydrologic conditions controlling runoff generation immediately after wildfire. Water Resources Research 48: W03529.

42.

Elliot

Robichaud

Hall

, et al. (2001) A probabilistic approach to modeling erosion for spatially varied conditions. ASAE paper number 01-8006, presented at the 2001 ASAE Annual International Meeting, Sacramento, 30 July–1 August, 16 pp.

43.

Elliott

Smith

Friedel

. (2004) Analysis and mapping of post-fire hydrologic hazards for the 2002 Hayman, Coal Seam, and Missionary Ridge wildfires. Scientific Investigations Report 2004-5300. Reston, VA: US Geological Survey, 104 pp.

44.

Emelko

Silins

Bladon

. (2011) Implications of land disturbance on drinking water treatability in a changing climate: Demonstrating the need for ‘source water supply and protection’ strategies. Water Research 45: 461–472.

45.

Fernández

Vega

Vieira

DCS

(2010) Assessing soil erosion after fire and rehabilitation treatments in NW Spain: Performance of RUSLE and revised Morgan-Morgan-Finney models. Land Degradation and Development 21: 58–67.

46.

Finney

Seli

McHugh

. (2007) Simulation of long-term landscape-level fuel treatment effects on large wildfires. International Journal of Wildland Fire 16: 712–727.

47.

Fiorucci

Gaetani

Minciardi

. (2008) Regional partitioning for wildfire regime characterization. Journal of Geophysical Research: Earth Surface 113: F02013.

48.

Flanagan

Gilley

Franti

(2007) Water Erosion Prediction Project (WEPP): Development history, model capabilities, and future enhancements. Transactions of the Asabe 50: 1603–1612.

49.

Gabet

Sternberg

(2008) The effects of vegetative ash on infiltration capacity, sediment transport, and the generation of progressively bulked debris flows. Geomorphology 101: 666–673.

50.

Gartner

Cannon

Santi

. (2008) Empirical models to predict the volumes of debris flows generated by recently burned basins in the western US. Geomorphology 96: 339–354.

51.

Gill

Allan

(2008) Large fires, fire effects and the fire-regime concept. International Journal of Wildland Fire 17: 688–695.

52.

Goode

Luce

Buffington

(2012) Enhanced sediment delivery in a changing climate in semi-arid mountain basins: Implications for water resource management and aquatic habitat in the northern Rocky Mountains. Geomorphology 139–140: 1–15.

53.

Gould

McCaw

Cheney

. (2007) Project Vesta. Fire in Dry Eucalypt Forest: Fuel Structure, Fuel Dynamics and Fire Behaviour. Perth: Ensis-CSIRO and Department of Environment and Conservation.

54.

Haimes

(2009) On the complex definition of risk: A systems-based approach. Risk Analysis: An International Journal 29: 1647–1654.

55.

Hennessy

Gregory

Mitchell

JFB

(1997) Changes in daily precipitation under enhanced greenhouse conditions. Climate Dynamics 13: 667–680.

56.

Hennessy

Lucas

Nicholls

. (2005) Climate change impacts on fire-weather in south-east Australia. Report. Aspendale: CSIRO Marine and Atmospheric Research.

57.

Hungerford

Harrington

Frandsen

. (1991) Influence of fire on factors that affect site productivity. Proceedings – Management and Productivity of Western-Montane Forest Soils 280: 32–50.

58.

Hyde

Dickinson

Bohrer

. (2012) Research and development supporting risk-based wildfire effects prediction for fuels and fire management: Status and needs. International Journal of Wildland Fire: 22: 37–50.

59.

Hyde

Woods

Donahue

(2007) Predicting gully rejuvenation after wildfire using remotely sensed burn severity data. Geomorphology 86: 496–511.

60.

Istanbulluoglu

Tarboton

Pack

. (2002) A probabilistic approach for channel initiation. Water Resources Research 38: 1325.

61.

Istanbulluoglu

Tarboton

Pack

. (2003) A sediment transport model for incision of gullies on steep topography. Water Resources Research 39: 1103.

62.

Istanbulluoglu

Tarboton

Pack

, et al. (2004) Modeling of the interactions between forest vegetation, disturbances, and sediment yields. Journal of Geophysical Research 109: F01009.

63.

Jakeman

Hornberger

(1993) How much complexity is warranted in a rainfall-runoff model? Water Resources Research 29: 2637–2649.

64.

Johnson

Gutsell

(1994) Fire frequency models, methods and interpretations. In: Begon

Fitter

(eds) Advances in Ecological Research. Amsterdam: Academic Press, 239–287.

65.

Jones

Nyman

Sheridan

(2011) A stochastic coverage model for erosion events caused by the intersection of burnt forest and convective thunderstorms. In: Chan

Marinova

Anderssen

(eds) MODSIM2011, 19th International Congress on Modelling and Simulation, Perth, Australia, 12–16 December, 1865–1871.

66.

Jorgensen

Hanshaw

Schmidt

. (2011) Value of a dual-polarised gap-filling radar in support of Southern California post-fire debris flow warnings. Journal of Hydrometeorology 12: 1581–1595.

67.

Julien

Lan

(1991) Rheology of hyperconcentrations. Journal of Hydraulic Engineering 117: 346–353.

68.

Kean

Staley

Cannon

(2011) In situ measurements of post-fire debris flows in southern California: Comparisons of the timing and magnitude of 24 debris-flow events with rainfall and soil moisture conditions. Journal of Geophysical Research – Earth Surface 116: F04019.

69.

Keane

Cary

Davies

. (2004) A classification of landscape fire succession models: Spatial simulations of fire and vegetation dynamics. Ecological Modelling 179: 3–27.

70.

Keane

Cary

Parsons

(2003) Using simulation to map fire regimes: An evaluation of approaches, strategies, and limitations. International Journal of Wildland Fire 12: 309–322.

71.

Keeley

(2009) Fire intensity, fire severity and burn severity: A brief review and suggested usage. International Journal of Wildland Fire 18: 116–126.

72.

Key

Benson

(2005) Landscape assessment: Remote sensing of severity, the normalized burn ratio and ground measure of severity, the composite burn index. In: Lutes

Keane

Caratti

, et al. (eds) FIREMON: Fire Effects Monitoring and Inventory System: General Tecnichal Report RMRS-GTR-164-CD . Ogden, UT: USDA Forest Service Rocky Mountain Research Station, LA1–LA51.

73.

King

Cary

Bradstock

. (2006) Simulation of prescribed burning strategies in south-west Tasmania, Australia: Effects on unplanned fires, fire regimes, and ecological management values. International Journal of Wildland Fire 15: 527–540.

74.

Kirchner

Finkel

Riebe

. (2001) Mountain erosion over 10 yr, 10 k.y., and 10 m.y. time scales. Geology 29: 591–594.

75.

Krajewski

Villarini

Smith

(2010) Radar-rainfall uncertainties. Bulletin of the American Meteorological Society 91: 87–94.

76.

Lane

PNJ

Sheridan

Noske

(2006) Changes in sediment loads and discharge from small mountain catchments following wildfire in south Eastern Australia. Journal of Hydrology 331: 495–510.

77.

Larsen

MacDonald

(2007) Predicting postfire sediment yields at the hillslope scale: Testing RUSLE and Disturbed WEPP. Water Resources Research 43: W11412.

78.

Barclay

Liu

. (2005) Simulation of historical and current fire regimes in central Saskatchewan. Forest Ecology and Management 208: 319–329.

79.

Liu

Gupta

Springer

. (2008) Linking science with environmental decision making: Experiences from an integrated modeling approach to supporting sustainable water resources management. Environmental Modelling and Software 23: 846–858.

80.

Livingston

Earles

Wright

(2005) Los Alamos post-fire watershed recovery: A curve-number-based evaluation. In: Moglen

(ed.) Managing Watersheds for Human and Natural Impacts. Alexandria, VA: American Society of Civil Engineers, 471–481.

81.

Luce

(2005) Land use and land cover effects on runoff processes: Fire. In: Anderson

(ed.) Encyclopedia of Hydrological Sciences. Chichester: Wiley, 1831–1838.

82.

Luce

Rieman

(2010) Landscape scale effects of fuel management or fire on water resources: The future of cumulative effects analysis? In: Elliot

Miller

Audin

(eds) Cumulative Watershed Effects of Fuel Management in the Western United States. Fort Collins, CO: US Department of Agriculture, Forest Service, Rocky Mountain Research Station, 234–245.

83.

Luce

Morgan

Dwire

. (2012) Climate change, forests, fire, water, and fish: Building resilient landscapes, streams, and managers. General Technical Report RMRS-GTR-290. Fort Collins, CO: US Department of Agriculture, Forest Service, Rocky Mountain Research Station, 207 pp.

84.

Malamud

Millington

JDA

Perry

GLW

. (2005) Characterizing wildfire regimes in the United States. Proceedings of the National Academy of Sciences of the United States of America 102: 4694–4699.

85.

Malamud

Morein

Turcotte

(1998) Forest fires: An example of self-organized critical behavior. Science 281: 1840–1842.

86.

Massman

Frank

Mooney

(2010) Advancing investigation and physical modeling of first-order fire effects in soils. Fire Ecology 6: 36–54.

87.

Merritt

Letcher

Jakeman

(2003) A review of erosion and sediment transport models. Environmental Modelling and Software 18: 761–799.

88.

Meyer

Pierce

Wood

. (2001) Fire, storms, and erosional events in the Idaho batholith. Hydrological Processes 15: 3025–3038.

89.

Meyer

Wells

Balling

. (1992) Response of alluvial systems to fire and climate change in Yellowstone National Park. Nature 357: 147–150.

90.

Michaud

Sorooshian

(1994) Comparison of simple versus complex distributed runoff models on a midsized semiarid watershed. Water Resources Research 30: 593–605.

91.

Miller

Luce

Benda

(2003a) Time, space, and episodicity of physical disturbance in streams. Forest Ecology and Management 178: 121–140.

92.

Miller

Nyhan

Yool

(2003b) Modeling potential erosion due to the Cerro Grande Fire with a GIS-based implementation of the Revised Universal Soil Loss Equation. International Journal of Wildland Fire 12: 85–100.

93.

Miller

MacDonald

Robichaud

. (2011) Predicting post-fire hillslope erosion in forest lands of the western United States. International Journal of Wildland Fire 20: 982–999.

94.

Moody

(2012) An analytical method for predicting postwildfire peak discharges. Scientific Investigations Report 2011-5236. Reston, VA: US Geological Survey, 36 pp.

95.

Moody

Ebel

(2013) Infiltration and runoff generation processes in fire-affected soils. Hydrological Processes. doi: 10.1002/hyp.9857.

96.

Moody

Martin

(2001a) Initial hydrologic and geomorphic response following a wildfire in the Colorado Front Range. Earth Surface Processes and Landforms 26: 1049–1070.

97.

Moody

Martin

(2001b) Post-fire, rainfall intensity-peak discharge relations for three mountainous watersheds in the western USA. Hydrological Processes 15: 2981–2993.

98.

Moody

Martin

(2009) Forest fire effects on geomorphic processes. In: Cerda

Robichaud

(eds) Fire Effects on Soils and Restoration Strategies. Enfield, NH: Science Publisher, 41–79.

99.

Moody

Martin

Haire

. (2008) Linking runoff response to burn severity after a wildfire. Hydrological Processes 22: 2063–2074.

100.

Moody

Shakesby

Robichaud

. (2013) Current research issues related to post-wildfire runoff and erosion processes. Earth-Science Reviews 122: 10–37.

101.

Moritz

Morais

Summerell

. (2005) Wildfires, complexity, and highly optimized tolerance. Proceedings of the National Academy of Sciences of the United States of America 102: 17912–17917.

102.

Moritz

Parisien

M-A

Batllori

, et al. (2012) Climate change and disruptions to global fire activity. Ecosphere 3: art49.

103.

Mouillot

Rambal

Joffre

(2002) Simulating climate change impacts on fire frequency and vegetation dynamics in a Mediterranean-type ecosystem. Global Change Biology 8: 423–437.

104.

Myronidis

Emmanouloudis

Mitsopoulos

. (2010) Soil erosion potential after fire and rehabilitation treatments in Greece. Environmental Modeling and Assessment 15: 239–250.

105.

Neary

Ryan

DeBano

(2005) Wildland fire in ecosystems: Effects of fire on soils and water. General Technical Report RMRS-GTR-42-vol 4. Fort Collins, CO: US Department of Agriculture, Forest Service, Rocky Mountain Research Station, 250 pp.

106.

Nott

(2006) Extreme Events: A Physical Reconstruction and Risk Assessment. Cambridge: UK: Cambridge University Press.

107.

Nyman

Sheridan

Lane

PNJ

(2010) Synergistic effects of water repellency and macropore flow on the hydraulic conductivity of a burned forest soil, south-east Australia. Hydrological Processes 24: 2871–2887.

108.

Nyman

Sheridan

Smith

. (2011) Evidence of debris flow occurrence after wildfire in upland catchments of south-east Australia. Geomorphology 125: 383–401.

109.

O’Brien

(2000) FLO-2D: Two-dimensional flood routing system. Reference manual, version 2009.06. Nutrioso, AZ: FLO-2D Software Inc., CD-ROM.

110.

O’Brien

Julien

Fullerton

(1993) Two-dimensional water flood and mudflow simulation. Journal of Hydraulic Engineering 119: 244–261.

111.

Pak

Lee

(2008) A statistical sediment yield prediction model incorporating the effect of fires and subsequent storm events. Journal of the American Water Resources Association 44: 689–699.

112.

Parisien

M-A

Moritz

(2009) Environmental controls on the distribution of wildfire at multiple spatial scales. Ecological Monographs 79: 127–154.

113.

Parsons

Robichaud

Lewis

. (2010) Field guide for mapping post-fire soil burn severity. General Technical Report RMRS-GTR-243. Fort Collins, CO: US Department of Agriculture, Forest Service, Rocky Mountain Research Station, 49 pp.

114.

Peterson

Moritz

Morais

. (2011) Modelling long-term fire regimes of southern California shrublands. International Journal of Wildland Fire 20: 1–16.

115.

Phillips

(2009) Changes, perturbations, and responses in geomorphic systems. Progress in Physical Geography 33: 17–30.

116.

Pierce

Meyer

Jull

AJT

(2004) Fire-induced erosion and millennial-scale climate change in northern ponderosa pine forests. Nature 432: 87–90.

117.

Piñol

Beven

Viegas

(2005) Modelling the effect of fire-exclusion and prescribed fire on wildfire size in Mediterranean ecosystems. Ecological Modelling 183: 397–409.

118.

Piñol

Terradas

Lloret

(1998) Climate warming, wildfire hazard, and wildfire occurrence in coastal eastern Spain. Climatic Change 38: 345–357.

119.

Podur

Martell

Stanford

(2010) A compound Poisson model for the annual area burned by forest fires in the province of Ontario. Environmetrics 21: 457–469.

120.

Price

Bradstock

(2011) Quantifying the influence of fuel age and weather on the annual extent of unplanned fires in the Sydney region of Australia. International Journal of Wildland Fire 20: 142–151.

121.

Price

Bradstock

Keeley

. (2012) The impact of antecedent fire area on burned area in southern California coastal ecosystems. Journal of Environmental Management 113: 301–307.

122.

Prosser

Williams

(1998) The effect of wildfire on runoff and erosion in native Eucalyptus forest. Hydrological Processes 12: 251–265.

123.

Reinhardt

Dickinson

(2010) First-order fire effects models for land management: Overview and issues. Fire Ecology 6: 131–142.

124.

Reinhardt

Keane

Brown

(2001) Modeling fire effects. International Journal of Wildland Fire 10: 373–380.

125.

Renard

Foster

Weesies

. (1991) RUSLE – Revised Universal Soil Loss Equation. Journal of Soil and Water Conservation 46: 30–33.

126.

Rickenmann

(1991) Hyperconcentrated flow and sediment transport at steep slopes. Journal of Hydraulic Engineering 117: 1419–1439.

127.

Ricotta

Avena

Marchetti

(1999) The flaming sandpile: Self-organized criticality and wildfires. Ecological Modelling 119: 73–77.

128.

Rieman

Lee

Burns

. (2003) Status of native fishes in the western United States and issues for fire and fuels management. Forest Ecology and Management 178: 197–211.

129.

Robichaud

(2005) Measurement of post-fire hillslope erosion to evaluate and model rehabilitation treatment effectiveness and recovery. International Journal of Wildland Fire 14: 475–485.

130.

Robichaud

Ashmun

(2012) Tools to aid post-wildfire assessment and erosion-mitigation treatment decisions. International Journal of Wildland Fire 22: 95–105.

131.

Robichaud

Elliot

Pierson

. (2007) Predicting postfire erosion and mitigation effectiveness with a web-based probabilistic erosion model. CATENA 71: 229–241.

132.

Robichaud

Wagenbrenner

Brown

(2010) Rill erosion in natural and disturbed forests: 1. Measurements. Water Resources Research 46: W10506.

133.

Rodriguez-Iturbe

Cox

Eagleson

(1986) Spatial modelling of total storm rainfall. Proceedings of the Royal Society of London A: Mathematical and Physical Sciences 403: 27–50.

134.

Rozalis

Morin

Yair

. (2010) Flash flood prediction using an uncalibrated hydrological model and radar rainfall data in a Mediterranean watershed under changing hydrological conditions. Journal of Hydrology 394: 245–255.

135.

Scott

Van Wyk

(1990) The effects of wildfire on soil wettability and hydrological behaviour of an afforested catchment. Journal of Hydrology 121: 239–256.

136.

Seed

Siriwardena

Sun

. (2002) On the calibration of Australian weather radars. Technical report to the Cooperative Research Centre for Catchment Hydrology no. 02/7. Monash University, Australia.

137.

Shakesby

Doerr

(2006) Wildfire as a hydrological and geomorphological agent. Earth-Science Reviews 74: 269–307.

138.

Sheridan

Lane

PNJ

Noske

(2007) Quantification of hillslope runoff and erosion processes before and after wildfire in a wet Eucalyptus forest. Journal of Hydrology 343: 12–28.

139.

Sheridan

Lane

Smith

. (2009) A rapid risk assessment procedure for post-fire hydrologic hazards: 2009/10 fire season. Technical Report to the Victorian Department of Sustainability and Environment (DSE). University of Melbourne, Australia.

140.

Sivakumar

(2008) Dominant processes concept, model simplification and classification framework in catchment hydrology. Stochastic Environmental Research and Risk Assessment 22: 737–748.

141.

Sivakumar

(2011) Global climate change and its impacts on water resources planning and management: Assessment and challenges. Stochastic Environmental Research and Risk Assessment 25: 583–600.

142.

Sivapalan

Blöschl

(1998) Transformation of point rainfall to areal rainfall: Intensity-duration-frequency curves. Journal of Hydrology 204: 150–167.

143.

Smith

Sheridan

Lane

PNJ

. (2011) Wildfire effects on water quality in forest catchments: A review with implications for water supply. Journal of Hydrology 396: 170–192.

144.

Smith

Sheridan

Nyman

, (2009) A framework for modelling suspended sediment flux following wildfire in forested water supply catchments, south-eastern Australia. In: Anderssen

Braddock

Newham

LTH

(eds) 18th World IMACS Congress and MODSIM2009 International Congress on Modelling and Simulation, Cairns, Australia, 13–17 July, 3549–3555.

145.

Springer

Hawkins

(2005) Curve number and peakflow responses following the Cerro Grande fire on a small watershed. In: Moglen

(ed.) Managing Watersheds for Human and Natural Impacts. Alexandria, VA: American Society of Civil Engineers, 459–470.

146.

Stephens

McIver

Boerner

REJ

. (2012) The effects of forest fuel-reduction Ttreatments in the United States. Bioscience 62: 549–560.

147.

Svensson

Jones

(2010) Review of rainfall frequency estimation methods. Journal of Flood Risk Management 3: 296–313.

148.

Swanson

(1981) Fires and geomorphic processes. In: Mooney

Bonnicksen

Christensen

. (eds) Fire and Ecosystem Processes. General Technical Report WO-26. Washington, DC: US Forest Service, 401–420.

149.

Syphard

Scheller

Ward

. (2011) Simulating landscape-scale effects of fuels treatments in the Sierra Nevada, California, USA. International Journal of Wildland Fire 20: 364–383.

150.

Tanaka

Kimura

Suga

(1983) Preparation of a 1:25000 Landsat map for assessment of burnt area on Etajima Island. International Journal of Remote Sensing 4: 17–31.

151.

Thompson

Calkin

(2011) Uncertainty and risk in wildland fire management: A review. Journal of Environmental Management 92: 1895–1909.

152.

Thompson

Calkin

Finney

. (2011) Integrated national-scale assessment of wildfire risk to human and ecological values. Stochastic Environmental Research and Risk Assessment 25: 1–20.

153.

Tomkins

Humphreys

Wilkinson

. (2007) Contemporary versus long-term denudation along a passive plate margin: The role of extreme events. Earth Surface Processes and Landforms 32: 1013–1031.

154.

Turner

Romme

(1994) Landscape dynamics in crown fire ecosystems. Landscape Ecology 9: 59–77.

155.

Vafeidis

Drake

Wainwright

(2007) A proposed method for modelling the hydrologic response of catchments to burning with the use of remote sensing and GIS. CATENA 70: 396–409.

156.

Victorian Department of Sustainability and Environment (2009) Remote sensing guideline for assessing landscape scale fire severity in Victoria’s forest estate. Reference manual for standard operating procedure No.4: Classification of remotely sensed imagery into fire severity. Melbourne: Department of Sustainability and Environment, Victoria.

157.

Vilén

Fernandes

(2011) Forest fires in Mediterranean countries: Emissions and mitigation possibilities through prescribed burning. Environmental Management 48: 558–567.

158.

Wade

Lunsford

(1989) A guide for prescribed burning in Southern forests. Technical Report R8-TP. Atlanta, GA: US Department of Agriculture, Forest Service, Southern Region, 56 pp.

159.

Westerling

Hidalgo

Cayan

. (2006) Warming and earlier spring increase western US forest wildfire activity. Science 313: 940–943.

160.

Williams

AAJ

Karoly

Tapper

(2001) The sensitivity of Australian Fire Danger to climate change. Climatic Change 49: 171–191.

161.

Wilson

Winter

Maguire

. (2011) Managing wildfire events: Risk-based decision making among a group of federal fire managers. Risk Analysis 31: 805–818.

162.

Wohl

Pearthree

(1991) Debris flows as geomorphic agents in the Huachuca Mountains of southeastern Arizona. Geomorphology 4: 273–292.

163.

Wondzell

King

(2003) Postfire erosional processes in the Pacific Northwest and Rocky Mountain regions. Forest Ecology and Management 178: 75–87.

164.

Wooster

Roberts

Perry

GLW

. (2005) Retrieval of biomass combustion rates and totals from fire radiative power observations: FRP derivation and calibration relationships between biomass consumption and fire radiative energy release. Journal of Geophysical Research: Atmospheres 110: D24311.